bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 High rates of visualized in the symbiont-bearing gills

2 of deep-sea Bathymodiolus mussels

3Piquet Bérénice1,2, Shillito Bruce2, Lallier François H.1, Duperron Sébastien2,3,4, Andersen Ann C.1

4

51: Sorbonne Université, CNRS, UMR7144, Lab. Adaptation et Diversité en Milieu Marin, AD2M, Team : Adaptation et

6Biologie des Invertébrés marins en Conditions Extrêmes, ABICE, Station Biologique de Roscoff, SBR, 29680 Roscoff,

7France.

82: Sorbonne Université, MNHN, CNRS, IRD, UCN, UA, UMR7208, Lab. Biologie des Organismes et Ecosystèmes

9Aquatiques BOREA, Team : Adaptation aux Milieux Extrêmes, AMEX, 7 Quai Saint-Bernard, 75005 Paris, France.

103: Muséum National d’Histoire Naturelle, CNRS, UMR7245, Lab. Mécanismes de Communication et Adaptation des

11Micro-organismes, Team : Cyanobactéries, Cyanotoxines et Environnement, CCE, 12 rue Buffon, 75005 Paris, France.

124: Institut Universitaire de France, Paris, France.

13

14Corresponding authors: [email protected] and [email protected]

1 1 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

15Abstract

16Symbiosis between Bathymodiolus and Gammaproteobacteria enables these deep-sea mussels to live in toxic

17environments like hydrothermal vents and cold seeps. The quantity of within the gill-

18bacteriocytes appears to vary according to the hosts environment. We investigated the hypothesis of a control

19of the endosymbionts density by apoptosis, a programmed cell death. We used fluorometric TUNEL-method

20and active Caspase-3-targeting antibodies to visualize and quantify apoptotic cells in mussel gills. To avoid

21artefacts due to depressurization upon specimen recovery from the deep-sea, we compared the apoptotic rates

22between mussels recovered unpressurised, versus mussels recovered in a pressure-maintaining device, in two

23species from hydrothermal vents on the Mid-Atlantic Ridge: Bathymodiolus azoricus and B. puteoserpentis.

24Our results show that pressurized recovery had no significant effect on the apoptotic rate in the gill filaments.

25Apoptotic levels were highest in the ciliated zone and in the circulating hemocytes, compared to the

26bacteriocyte zone. Apoptotic gill-cells in B. aff. boomerang from the pockmarks off the Gulf of Guinea,

27show similar distribution patterns. Deep-sea symbiotic mussels have much higher rates of apoptosis in their

28gills than the coastal mussel Mytilus edulis without chemolithoautotrophic symbionts. We discuss how

29apoptosis might be one of the mechanisms that contribute to the adaptation of deep-sea mussels to toxic

30environments and/or to symbiosis.

31

32Key words: symbiosis, isobaric recovery, hydrothermal vents, cold seeps, Mytilus edulis, bacteriocytes.

2 2 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

33Introduction

34 Symbiosis is of major significance to life on Earth. Because symbiosis is a mixture of cooperation and

35conflict between two (or several) partners, each of them must initiate and carry on a continued dialogue, and

36control their partners’ interaction. Many symbioses involve partners from different domains of life, such as a

37eukaryote host and a bacterial symbiont. This questions how a reciprocal control can occur between so

38distantly related organisms, and whether some of the mechanisms involved might be universal. One of the

39mechanisms by which an animal host can control populations of its symbionts is apoptosis. Apoptosis is a

40programmed cell death involving three main steps: (1) nuclear condensation and fragmentation, (2) cell-wall

41budding into apoptotic bodies, and (3) their release and possible phagocytosis by neighboring cells [1,2].

42Apoptosis plays multiple roles in normal cell turnover, during development, and in the immune system [3].

43Its role in symbiosis is also documented: in the cereal weevil Sitophilus for example, apoptosis participates to

44regulate the densities of the endosymbiotic bacterium Sodalis pierantonius. The symbiont provides essential

45amino acids that allow the host to rapidly build its protective exoskeleton. Endosymbionts multiply in young

46adults, but when the cuticle is built, the symbionts are rapidly eliminated through apoptosis of the host cells

47that contain them [4]. In corals, bleaching occurs in response to heat stress, when the host releases

48dinoflagellate symbionts through apoptosis and linked in a see-saw manner, such that when

49apoptosis is inhibited autophagy is initiated as a back-up mechanism, and vice-versa [5]. In addition

50apoptosis might act as a post-phagocytic winnowing mechanism in the symbiotic system [6].

51 At deep-sea hydrothermal vents and cold seeps, the animals that dominate in terms of biomass live in

52association with chemosynthetic , which sustain most of their nutrition [7–10], but the role apoptosis

53could play in regulating symbiosis has barely been explored. In the vestimentiferan tubeworms Riftia

54pachyptila, living at vents, and Lamellibrachia luymesi from cold seeps, the sulfur-oxidizing symbionts are

3 3 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

55located within cells of the . These cells differentiate and proliferate from the trophosome lobule

56center, then migrate towards the periphery of the lobule where they undergo apoptosis [11]. Ultrastructural

57observations at the periphery of the trophosome lobules show that the symbionts are digested in

58leading to extensive myelin bodies, then remnants of symbionts disappear, while apoptotic nuclei with

59clumped chromatin patches occur [11]. Thus, in Riftia as in the weevil, apoptosis appears to be involved in

60the process of symbiont regulation to recover the metabolic investment from the symbiotic phase.

61 Deep-sea mussels house very dense populations of endosymbionts inside specialized gill epithelial

62cells, the bacteriocytes, [12,13]. In fact, Bathymodiolus constitute by far the densest microbial habitats,

63although they usually host a very limited diversity of symbiont lineages [7,8]. The relevance to the topic of

64symbiont control lies in the fact that the association is particularly flexible, with abundances of their

65symbionts (Sulfur- and/or methane-oxidizers) that can vary within hours depending on the availability of

66symbiont substrates in the surrounding water [8,14–18]. The symbionts also rapidly disappear if their

67substrates are absent [17–19]. Ultrastructural studies of the gill cells have indicated intracellular digestion of

68the symbionts within as an important carbon transfer mechanism [20–22] suggesting that the host

69can access the energy stored in its symbionts by killing and digesting them (i.e. a process compared to

70“farming”). Enzymatic studies involving the detection of acid phosphatases have shown that some energy

71from the symbionts can also be transferred to the host through molecules leaking from live symbionts (i.e. a

72process compared to “milking”) [23].Recent results from whole-gill tissue transcriptomic analyses in the

73vent species Bathymodiolus thermophilus indicated that high symbiont loads are correlated with under-

74expression of the genes inhibiting apoptosis, suggesting that when the symbionts are abundant, apoptosis

75might be activated [24,25]. It can thus be hypothesized that apoptosis is a mechanism by which the hosts

76control the number of symbionts inside their gills, and possibly obtain their carbon. High throughput

4 4 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

77sequencing and transcriptomic analyses have shown the great importance of the apoptotic signaling pathways

78in the gill tissue of several species of Bathymodiolus [26–28].

79 Apoptosis in mollusks is generally triggered by the interaction between immune cells and parasites

80or pathogens [29]. The high degree of evolutionary conservation of biochemical signaling and executing

81pathways of apoptosis indicate that programmed cell death likely plays a crucial role in homeostasis and

82functioning of molluscan immune system [29–31]. Apoptosis is particularly complex in mollusks, and

83caspases are key molecules that are activated in two major apoptotic pathways: the extrinsic or death

84receptor pathway, and the intrinsic or mitochondrial pathway [31]. The executioner caspase-3 activates a

85heterodimer , the DNA fragmentation factor that is responsible for the completion of DNA

86fragmentation [31]. However alternative caspase-independent pathways have also been evidenced in

87mollusks, and cross-talk between different pathways might also be involved [30,31]. Apoptosis was shown to

88be induced in Bathymodiolus azoricus in response to Vibrio diabolicus exposure [32]. However the immune

89gene responses in B. azoricus appeared tied to the presence of bacteria, as the progressive

90weakening of its host transcriptional activity correlates with the gradual disappearance of endosymbiont

91bacteria from the gill tissues during the extended acclimatization in the sulfide and methane-free aquaria

92[32]. Thus, the presence of symbionts might modulate the apoptotic patterns in deep-sea symbiotic mussels

93compared to their coastal mussel relatives without chemolithoautotrophic symbiont.

94 While several studies have highlighted an important activity of apoptotic signaling factors in three

95species of Bathymodiolus (B. azoricus, B. manusensis from hydrothermal vents, and B. platifrons from cold

96seeps (Bettencourt et al., 2010; Wong et al., 2015; Zheng et al., 2017)), transcriptomic studies do not give a

97visual account of this apoptotic activity within the tissues. The present study is the first microscopic

98investigation of the general distribution patterns of apoptotic cells in the gills, and the first attempt to

5 5 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

99quantify apoptosis in the different gill cell types of Bathymodiolus. We chose to focus on three species,

100namely Bathymodiolus azoricus, B. puteoserpentis and B. aff. boomerang. The first two often dominate the

101macrofauna at Mid-Atlantic Ridge (MAR) hydrothermal vent sites [33,34]. The third occurs at cold seeps

102located on the continental margin in the Gulf of Guinea [35]. The former two species are phylogenetically

103sister species and are also closely related to B. boomerang [12]. All three species harbor methane- and sulfur-

104oxidizing bacteria that co-exist within host cells. Their gills comprise several cell types: epidermal ciliated

105cells and mucous goblet cells, bacteriocytes hosting the symbionts, interspaced by intercalary cells, and

106finally circulating hemocytes [36]. Our aim was to investigate, whether apoptosis might contribute to

107regulating symbiont densities. We thus tested, whether apoptosis preferentially occurs in bacteriocytes, in

108particular the most densely populated ones.

109 As the recovery of deep-sea specimens from several thousand meters depth usually involves a

110depressurization stress that might alter gene expression and disturb the cell machinery, leading to artefacts

111[37,38] we performed a comparative analysis of apoptosis in specimens recovered with and without

112depressurization upon collection. Specimens of B. azoricus and B. puteoserpentis were recovered using an

113hyperbaric chamber (PERISCOP - Projet d’Enceinte de Remontée Isobare Servant la Capture d’Organismes

114Profonds) that allows maintaining their native pressure and temperature throughout recovery, avoiding

115recovery bias [39]. Control specimens recovered without PERISCOP were included. The cold seep B. aff.

116boomerang was analyzed to identify potential seep versus vent habitat-related differences, and Mytilus edulis

117was used as a non-chemolithoautotrophic-symbiont-bearing control mussel. Apoptosis was visualized in gill

118tissue sections by the TUNEL method (Transferase dUTP Nick-End Labeling) [11,40,41], a common and

119standard way of visualizing the fragmented DNA in the nucleus of cells undergoing apoptosis. To further

120support that observed patterns actually correspond to apoptosis, we performed immunolocalization of active

6 6 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

121Caspase-3, a form of this enzyme that is the overall convergent node of molecular cascades leading to

122irreversible apoptosis in mollusks [31]. Altogether, this study provides to our knowledge the first visual

123overview of apoptosis in relation to symbiosis in chemosynthetic mussels.

124

125Materials and Methods

126Specimen collection

127 This analysis was conducted on 40 individuals of Bathymodiolus (Bivalvia, Mytilidae) consisting in

12821 Bathymodiolus azoricus and ten B. puteoserpentis from hydrothermal vents, and nine B. aff. boomerang

129from cold seeps. Specimens of B. azoricus were collected during the BioBaz 2013 cruise on the Mid-Atlantic

130Ridge [42]: ten specimens were sampled from the vent fields of Menez Gwen (MG2 marker: 37°50.669N;

13131°31.156W, 830m depth) and eleven specimens from Rainbow site (France 5 marker: 37°17.349N;

13232°16.536W, at 2270 m depth). B. puteoserpentis were sampled during the BICOSE 2014 cruise [43]. All

133ten specimens were sampled on the vent site Snake Pit, close to the “Elan” marker (23°22′54′′N, 44°55′48′

134′W, at 3520 m depth). For these two species, half of the specimens were recovered in a standard (i.e.

135unpressurized) way, and the other half using the PERISCOP hyperbaric vessel. The standard collection was

136done in a waterproof sealed box (BIOBOX) containing local deep-sea water, in which the mussels were

137exposed to a depressurization corresponding to the depth of their habitat during recovery (approx. 0.1 MPa

138every 10 m). The pressurized recovery was performed using a small device named the "Croco" sampling cell

139[44], which was transferred into the PERISCOP pressure-maintaining device [39]. PERISCOP was released

140from the seafloor through a shuttle system, and surfaced within 45 to 100 minutes. Pressure was monitored

7 7 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

141during ascent with an autonomous pressure sensor (SP2T4000, NKE Instruments, France). To compare

142apoptosis between vent and seep mussels, nine specimens of Bathymodiolus aff. boomerang were collected

143in a standard manner from the Régab pockmark site in the Gulf of Guinea (M1 marker 5°47.89S, 9°42.62E,

1443150 m depth and M2 marker 5°47.85S, 9°42.67E, 3150 m depth) during the 2011 cruise WACS [45]. All

145three cruises were aboard the RV Pourquoi Pas? using the ROV Victor 6000. In addition to deep-sea

146mussels, nine coastal mussels (Mytilus edulis) in two sets were analyzed. First set were wild M. edulis

147collected on rocks from the intertidal seashore in front of Roscoff Marine Station in January 2014 (n=4

148individuals analyzed). The second set of M. edulis were cultivated mussels bought from the fishmongers in

149Paris in January 2017 (n= 5 individuals analyzed). All data concerning the specimens are shown in the

150Supplementary Table 1.

151 Sample fixation, inclusion and FISH experiments

152 Mussel gills were dissected at 4°C, within 10 minutes after recovery (releasing pressure of the

153PERISCOP in the case of isobaric recoveries). Maximum shell length, height under the umbo, and width of

154the closed shell were measured with a caliper (see supplementary figure 1). Anterior and posterior gill

155fragments were fixed in 4% formaldehyde in sterile-filtered seawater (SFS) for 2 hours. Gills were then

156rinsed in SFS, and dehydrated in increasing series of ethanol (50, 70 and 80%, 15 min each). In the

157laboratory, gills were embedded in polyethylene glycol (PEG) distearate: 1-hexadecanol (9:1), cut into 8 µm-

158thick sections using a microtome (Thermo, Germany), recovered on SuperFrost Plus slides (VWR

159International, USA), and stored at -20°C. Fluorescence in situ hybridization (FISH) experiments were

160performed to confirm the localization of symbionts following the protocol and using probes described

161previously [15]. The gill lamellae were sectioned ventrally, so that each filament appeared in cross-section

8 8 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

162with its frontal end facing externally towards mantle and shell, and its abfrontal end facing the inner gill

163lamellae towards the foot. (See supplementary figure 2).

164TUNEL (Transferase dUTP Nick-End Labeling)

165 For the detection and quantification of apoptosis, we used the TUNEL method with the in-situ cell

166death detection kit following manufacturer’s instructions (ROCHE, Germany). All slides were first dewaxed

167then rehydrated by immersion in decreasing series of ethanol. Tissues were rinsed with PBS 1 X and

168permeabilized by proteinase K (20µg.ml-1) to enable the binding of dUTP. Various incubation times were

169tested, and 8 min gave the best result. Slides were incubated at 37°C for 1h30 with fluorophore dUTP and

170enzyme rTdT. Slides were rinsed in 3 PBS bathes (10 min each) to remove the unfixed fluorescein-12-dUTP.

171In a last step, DNA of all nuclei was labelled with 4’6’-Diamidino-2-phenylindole (DAPI) that is

172incorporated in the SlowFade® Gold Antifade Mounting reagent (Invitrogen, USA). For each analyzed

173individual, positive and negative controls were performed on adjacent serial sections. The positive control

174involved a pre-incubation with DNase I (3U/ml, 10 min) before running the full protocol. It provokes

175artificial fragmentation of DNA and exposes 3’OH ends, which bind the green fluorochrome and leads to all

176nuclei being labelled. The negative control follows the same protocol, but omitting the rTdT enzyme. It

177enables to reveal autofluorescence of the tissues and unspecific fixation of the fluorophore.

178Immunolocalization of active Caspase-3

179 Slides were dewaxed and rehydrated as for the TUNEL experiments. They were then covered in

180blocking solution (PBS 10X, 2% BSA, 0.3% triton) for 2 h at 4°C. Tissue sections were incubated with the

181rabbit polyclonal active anti-Caspase-3 primary antibody (directed against a peptide from the P17 fragment

182of the activated human/mouse Caspase-3) (R&D system, USA). This primary incubation lasted 2 hours at

9 9 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1834°C. Slides were rinsed in PBS 10X three times and covered with the secondary antibody: Alexa Fluor 488-

184labelled goat anti-rabbit (Invitrogen, dilution 1:500) for 1 h at room temperature. After 3 baths in PBS 10X,

185slides were mounted with SlowFade®. Negative controls were obtained by omitting the primary antibody.

186Image acquisition and analysis

187 Slides were observed under a BX61 epifluorescence microscope (Olympus, Japan), and pictures

188were taken under an SP5 confocal (Leica, Germany) at the x40 magnification. This magnification enabled a

189clear resolution to count each individual labeled nucleus. Entire cross-sections of the TUNEL-labelled gill

190filaments were covered in full in about 3 to 4 pictures (figures 1 and 2). The exposure time (or gain on the

191confocal) was standardized and identical for all pictures. The 3D-acquisitions were obtained by acquiring

192images every 0.5 µm throughout the thickness of the section. Slides were observed at 505 nm and 555 nm

193wavelengths (TUNEL and Caspase 3 signals, respectively), and 400 nm (DAPI). Classical immuno-

194localization are shown with all labeling in their original color (DAPI in blue), but for TUNEL labelled nuclei

195we had to change the DAPI color into red. Indeed blue DNA-labelling overlaid by green (i.e. apoptosis

196TUNEL labeling) leads to a complex mixture of various levels of green and blue, often making quantitative

197interpretations difficult. To enable an easy quantitative analysis of the TUNEL images, we chose to visualize

198DAPI in red, so that a double labelling would lead to a third distinct color (yellowish). Thus, a TUNEL-

199labelled nucleus appears bright green when its amount of fragmented DNA is great (i.e. when a nucleus is in

200an advanced state of apoptosis), but yellow or orange, when its amount of fragmented DNA is respectively

201low or very low (i.e. when a nucleus is in an early state of apoptosis). At first, we counted green and

202yellowish nuclei separately, but as they both expressed apoptosis, and early and advanced states were

203randomly distributed in the gill lamellae (figures 1 and 2), we finally pooled them together.

10 10 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

204Figure 1: FISH and TUNEL labelling on gill filaments of B. azoricus. A: Transverse section of gill filaments with FISH labelling. 205Nuclei from host tissue in blue (DAPI); sulfur-oxidizing symbionts in pink, and methane-oxidizers in green. B, C and D TUNEL- 206labelled nuclei in green, DAPI staining in red. Arrows point to apoptotic nuclei. B: Ciliated zones often contain many labelled nuclei. 207C: Bacteriocyte zone that displays many apoptotic nuclei (arrows), apoptotic hemocytes (arrowheads) and an apoptotic cell detached 208from the basal lamina (circle). D: Abfrontal zone of the gill filament showing apoptotic nuclei in cells with no symbiont (arrows), 209and apoptotic cells detached from the basal lamina (circles). Abbreviations: BZ: Bacteriocyte zone; FCZ: Frontal Ciliated Zone; H: 210Hemolymph zone; IFCJ Inter-Filamentary Ciliated Junction. Scale bars: 50µm. 211

212Figure 2: TUNEL labelling on gill filaments of B. puteoserpentis. TUNEL-labelled nuclei in green, DAPI staining in red. A: 213Frontal ciliated zones often contain many labelled nuclei. B: Bacteriocyte zone that displays only a few apoptotic nuclei. C: Inter 214Filamentary ciliated Junction often contain many labelled nuclei. D: Abfrontal zone of the gill filament with apoptotic nuclei. 215Abbreviations: BZ: Bacteriocyte zone; FCZ: Frontal Ciliated Zone; H: zone; IFCJ Inter-Filamentary Ciliated Junction. 216Scale bars: 50µm. 217

218 Images were analyzed by counting the labelled nuclei using the free Image J software [46]. However

219care was taken to count separately each respective cell-type: hemocytes, ciliated cells and finally

220bacteriocytes and intercalary cells (the two latter being close neighboring cell types, hard to distinguish

221under the fluorescence microscope, they were counted together) (figure 1 and 2). Thus for each image,

222percentages of TUNEL-labelled nuclei were obtained from the hemolymph zone (HZ), the ciliated zone

223(CZ), and the bacteriocyte zone (BZ) by calculating the ratio between the number of TUNEL-labelled nuclei

224and the number of DAPI-labeled nuclei present in each respective zone. A fourth value, the total percentage

225for the whole picture, was obtained by summing all labeled nuclei (whatever their cell-type) on the total

226number of DAPI-stained nuclei in a given picture.

227Comparisons between species, zones and treatments

228 The percentage of TUNEL nuclei labelled was used for all analyses after an Arcsine transformation

229[47]. The normality of datasets was tested (Shapiro-Wilk test), which revealed a non-normal distribution of

230the data. Non-parametric tests were thus applied for inter-groups comparisons (Mann–Whitney–Wilcoxon

231and Kruskal–Wallis) with the Bonferroni correction for multiple comparisons. All statistical analyses and

232graph plots were performed using R (R Development Core Team, version 3.3.3).

11 11 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

233

234Results

235 All analyses were performed on transverse sections of the gill lamellae (supplementary figure 2).

236From the FISH results, it is noteworthy that bacteriocytes close to the frontal ciliated zone contained large

237quantities of endosymbionts, and that both the height of the bacteriocytes and their symbiont density

238decreased towards the abfrontal zone in B. puteoserpentis, and B. azoricus (figure 1A). This frontal/abfrontal

239decrease in symbiont density also appears with the DAPI staining, as this labels not only the host nuclei, but

240also the DNA of the bacterial symbionts.

241Visualization of apoptosis in Bathymodiolus

242 The distribution of TUNEL positive cells in B. azoricus and B. puteoserpentis are shown in Fig. 1

243and 2, respectively. In negative controls (supplementary figure 3B), no fluorescence at all was observed

244compared to positive controls (supplementary figure 3A), except a bright auto-fluorescence signal seen in

245clustered round granules. These granules are thought to correspond to mucus droplets present in goblet-cells

246interspersed among the ciliated cells and along the epithelium of the gill filaments. This autofluorescence

247signal was easy to distinguish from TUNEL signal, as the latter was typically much smaller and brighter.

248Nuclei labelled in the TUNEL experiments were more or less bright. As TUNEL labels the free 3’OH

249fragments, we assume that the strongly labelled nuclei showed a higher amount of DNA fragments compared

250to the weakly stained ones. Since apoptosis is a progressive process occurring within a few hours, we

251hypothesize that the former were in a more advanced apoptotic state. Both weak (orange or even yellow) and

12 12 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

252strong-labelled nuclei (green) were thus counted together to estimate the percentage of apoptotic nuclei in the

253various cell-types of the gill lamellae (ciliated zone, bacteriocyte zone and hemolymph) (figure 1and 2).

254Figure 3: Active Caspase-3 labelling in the gill of B. azoricus. Typical labelling in green surrounds the nucleus (arrows). DAPI 255signal in blue. Scale bar: 50µm. 256 257 The ciliated frontal zone often is the brightest labelled zone with many strongly labelled cells: up to

258one cell out of two in some filaments (Fig. 1B, 2A) giving a clustered appearance of the apoptotic cells in the

259frontal ciliated zone. However the closest neighbor filament may sometimes only have a few or no labelled

260cells at all, indicating a great spatial variability in the tissue taken all together. Hemocytes were present on all

261images, but in varying numbers. Many hemocytes are visible on figure 1C for example, compared to figure

2621B, but more of them were labelled in the latter. TUNEL-labelling in the bacteriocyte zone was

263heterogeneous: very few nuclei were labelled close to the frontal ciliated zone (figure 1B and figure 2A), and

264then increasingly more were labelled when reaching the abfrontal zone (figure 1D and figure 2D). In the

265abfrontal zone, the bacteriocytes were most of the time devoid of bacteria. Sometimes, a bacteriocyte seemed

266to be lacking in the abfrontal zone, leaving a “hole” in the epithelial tissue (figure 1C-D and figure 2 B-D).

267 Active Caspase 3 immunohistochemistry assays yielded strong signals within the cytoplasm of cells,

268and particularly concentrated around the nuclei (figure 3). On serial sections within the same individual, the

269distribution patterns of active Caspase-3 were similar to the TUNEL signal observed, confirming that

270TUNEL actually revealed apoptotic cells. In general, the active Caspase-3 antibody seemed to label more

271nuclei than TUNEL.

272Patterns of apoptosis in Bathymodiolus azoricus and B. puteoserpentis

273 In total, 31 individuals from the two species were analyzed, representing a total of 19,612 DAPI-

274labelled nuclei counted on 206 images. The specimens came from deep-sea sites located at different depths,

13 13 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

275namely Menez Gwen (-830 m), Rainbow (-2270 m) and Snake Pit (-3520 m). Sampled specimens were

276either recovered unpressurised (non-isobaric) or with a pressurized (isobaric) recovery. In the latter,

277PERISCOP reached the surface while retaining 83.6, 76.5 and 70.4% of the native deep-sea pressure at the

278Menez Gwen, Rainbow and Snake Pit sites, respectively.

279 First of all, comparing the global counts in gills of all individuals from all three sites, we noted that

280there is a great variation in the percentage of apoptotic nuclei among individuals (see supplementary figure 4

281A-C). This strong heterogeneity indicates that results of statistical analyses must be treated with caution. In

282the ciliated (CZ) and bacteriocyte zones (BZ), the percentage of TUNEL-labelled nuclei was not

283significantly different among the three MAR sites and the two recovery types (CZ: Kruskal-Wallis test chi-

284squared =6.07, df=5, p-value=0.30; BZ: Kruskal-Wallis test chi-squared=3.69, df=5, p-value=0.60) (See

285figure 4 for BZ and supplementary figure 5 A for CZ). Similarly, the hemolymph zone showed no significant

286difference among sites and recovery types (HZ: Kruskal-Wallis test chi-squared=10.74, df=5, p-

287value=0.05678) (supplementary figure 5 B). These quantitative results confirm the visual patterns and

288intensities of TUNEL-labelling in gill filaments of mussels that did not reveal any self-evident difference

289between specimens from the three sites in either recovery type. For all specimens put together,

290quantifications indicated median values of 41.3% (±27%) and 34.5% (±31%) labelled nuclei in the ciliated

291and hemolymph zones, respectively. In comparison, the bacteriocyte zone displayed only 19.3% (±24%) of

292nuclei that were labelled, (figure 5) which is significantly less than in the other two (Kruskal-Wallis test chi-

293squared=27.639, df=2, p-value<0.001, post-hoc test with Bonferroni correction between HZ and BZ, p-

294value=0.0002; between CZ and BZ, p-value <0.0001).

295Figure 4: Percentages of apoptotic nuclei in the bacteriocyte zone of B. azoricus (Ba) and B. puteoserpentis (Bp) from the 296three sites. White boxplots indicate specimens from non-isobaric, and grey from isobaric recoveries, respectively. Percentages were 297not significantly different (a) in pairwise Wilcoxon test with Bonferroni’s standard correction. The letter n indicates the number of 298individuals analyzed for each species. Boxplot whiskers correspond to minimal and maximal values on a single image, the line inside 299the box is the median, and the upper and lower frames of the boxes represent the first and third quartile respectively. 300

14 14 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

301Figure 5: Percentages of apoptotic nuclei in the hemolymph, ciliated and bacteriocyte zones of B. azoricus and B. 302puteoserpentis. Different letters (a,b) indicate the plots that are statistically different (Wilcoxon test) and n indicates the number of 303specimens. Boxplot whiskers indicate minimal and maximal values on a single image, the line inside the box is the median, and the 304upper and lower frames of the boxes represent the first and third quartile respectively. 305

306Patterns of apoptosis in the seep species Bathymodiolus aff. boomerang

307 TUNEL labelling was applied to the gill filaments of Bathymodiolus aff. boomerang 9 specimens, 40

308pictures representing 3,424 nuclei) (figure 6 A). Percentages were significantly different between zones

309(Kruskal-wallis test: chi-squared=25.00, df=2, p-value<0.0001). As in the two studied vent species, apoptosis

310was more frequent in the ciliated zone (median: 36%, ±23%) and then in the hemocytes zone (median: 21%

311±9%), compared to the bacteriocyte zone (median: 9% ±23%). Overall these results suggest similar patterns

312in Mytilidae from hydrothermal vents and cold seeps (figure 7). Visually, the density of bacteria inside

313bacteriocytes seems to be more constant between bacteriocytes located in the frontal and in the abfrontal

314distal edge in B. aff. boomerang compared to the two vent mussels.

315Figure 6: TUNEL labelling on gill filaments of Bathymodiolus aff. boomerang (A) and Mytilus edulis (B-C). TUNEL-labelling 316in green or yellow, DAPI staining in red. A: Bathymodiolus aff. boomerang displays many labelled nuclei mostly in the frontal 317ciliated zone. B-C: Mytilus edulis gills show very few TUNEL-labelled nuclei, two are visible in the insert (C). Same abbreviations 318as in figure 1. Scale bar: 50µm. 319

320Figure 7: Percentages of apoptotic nuclei compared between vent and seep species in the ciliated, hemolymph and 321bacteriocyte zones of B. azoricus, B. puteoserpentis and B. aff. boomerang (Ba, Bp and Bb, respectively). The letter n indicates 322the number of specimens. Boxplot whiskers indicate minimal and maximal values on a single image, line inside the box is the 323median, the line inside the box is the median, and the upper and lower frames of the boxes represent the first and third quartile 324respectively. 325

326Patterns of apoptosis in the gill of coastal Mytilus edulis

327 Gill tissues of Mytilus edulis were labelled by the TUNEL method (9 specimens from two different

328localizations, 28 images, 4,091 nuclei). Both the wild mussels, collected on the intertidal rocks near the

329marine station in January and the commercial mussels bought in Paris in January, gave the same distribution

15 15 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

330pattern of apoptotic cells. The number of labelled nuclei was very low in comparison to vent and seep

331mussels (figure 6B) (median: 1.6%, ±1.6%) and nearly the same percentage between pictures. In M. edulis,

332the majority of labelled nuclei were hemocytes (median: 6.9%, ±10.9%) compared to the epidermal cells

333(median: 0.9%, ±1.4%).

334

335Discussion

336Relevance of TUNEL labelling to the study of apoptosis in deep-sea mussels

337 During apoptosis DNA undergoes fragmentation, and the TUNEL assay labels the free 3’OH ends

338that are generated. However, other mechanisms can lead to DNA fragmentation that would result in a

339positive TUNEL labelling, and for this reason the TUNEL methodology has been criticized [48–50]. It is

340nevertheless a standard procedure that has already been applied in mollusk tissues, and in association with

341active caspase-3 immunolocalization to detect apoptosis in the chemosymbiotic bivalve Codakia orbiculata

342[40,51]. Caspase-3 exists within the cytosol as inactive dimers. All apoptotic pathways in mollusks converge

343towards a step of cleavage of this zymogen that ultimately results in the triggering of apoptosis [31,52].

344Although it may have other roles, the activation of Caspase-3 usually leads to cell death by apoptosis.

345Because other caspases activate necrosis and necroptosis pathways, the occurrence of active Caspase-3

346signals rather supports that most observed TUNEL signals actually correspond to apoptotic nuclei [53]. It

347should however be kept in mind that another apoptotic pathway exists that is Caspase-independent, as it can

348take place without activating Caspase-3 [54,55].

349Deep-sea mussels display high rates of apoptosis in their gills

16 16 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

350 Recovery stress has long been cited as a factor that may prevent accurate assessment of the

351physiology of deep-sea organisms [39,56,57]. In this study, we found no statistical difference between

352percentages of apoptotic cells between gills from specimens of Bathymodiolus azoricus and B.

353puteoserpentis recovered classically (non-isobaric) and those recovered using the isobaric sampling cell

354PERISCOP. Thus depressurization during recovery does not seem to trigger massive apoptosis in the gills,

355despite that depressurization, by disrupting cells, may alter the distribution of different actors of apoptosis

356that pass through different cell compartments (mitochondria, nucleus, cell membrane) [39]. No differences

357were observed between hydrothermal vent sites, whatever their depth, further suggesting that apoptosis is a

358natural phenomenon in the gills of deep-sea mussels. Estimates of rates of apoptosis based on TUNEL

359labelling were much higher in the gills of deep-sea mussels including the seep species B. aff. boomerang,

360compared to their coastal relative Mytilus edulis lacking chemolithotrophic symbionts. This comparison has

361been made with two different sets of Mytilus: wild and cultivated, yielding the same scarce TUNEL-

362distribution pattern. At this stage it is not possible to ascertain, whether the much higher levels measured in

363Bathymodiolus are linked with their deep-sea chemosynthetic habitats (for example due to the abundance of

364xenobiotic compounds and oxidative stressors), with the occurrence of symbionts, or with their different

365evolutionary history.

366Patterns and potential roles of apoptosis in deep-sea mussels

367 A major finding in this study is that rates of apoptosis vary considerably among Bathymodiolus

368individual specimens, contrary to Mytilus, for a similar investigation effort. This could reveal distinct

369physiological status among the Bathymodiolus at the time of the analysis. Nonetheless, it should be

370underlined that apoptosis is known to be a rapidly occurring mechanism, in a time range of only a few hours

371[1]. The distribution of apoptotic cells in a tissue often shows a clustered pattern, as for instance in the

17 17 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

372mammalian intestine, where apoptotic signal appears typically confined to a cluster of neighboring villi,

373while other areas of the mucosa appear unstained [41]. A sporadic distribution might ensure a local recycling

374of the organ, without harming its global function, and might indeed be a characteristic of the apoptotic

375phenomenon. In the case of Bathymodiolus, gills often showed a few TUNEL-positive cells in some cross-

376sections, and even clusters in the frontal ciliated area of some filaments, but none in the frontal zone of

377neighboring filaments. This clustered pattern may also contribute to the observed heterogeneity among our

378individuals. Anyway, the great inter-individual variation prevents from drawing any definitive conclusion,

379even based on statistical tests of which the results have to be treated with caution. Nevertheless, a clear trend

380is that the rates measured in the bacteriocyte zone are lower than in the ciliated and hemocyte zones. The

381very high rates of apoptosis in the ciliated gill cells, in particular compared to M. edulis, are intriguing

382because these cells do not contain chemolithotrophic symbionts. The gills of Bathymodiolus are clearly

383different from those of the coastal mussels for which, at any given size, the gill surface area is around 20

384times smaller [8]. Bathymodiolus gills are particularly thick and over-developed, which might lead to a

385higher metabolic demand in order to properly circulate water and vital compounds. Ciliated cells also display

386very abundant mitochondria [8,13,58]. One hypothesis could thus be that the great metabolic activity and

387numerous mitochondria produce large amounts of (ROS), resulting in oxidative

388stress that may lead to increased apoptosis [59]. A possible adaptation would be to increase the turnover rates

389of the ciliated cells. To achieve this, high rates of cell proliferation could be expected, and this has to be

390tested. Another hypothesis is linked with fluid toxicity and the direction of water flow in the gills. Ciliated

391cells are indeed the first exposed to the surrounding fluid, and thus exposed to the highest levels of reduced

392compounds including toxic sulfur and various metals that possibly trigger apoptosis [60]. This may also

393explain why symbionts tend to be more abundant in the bacteriocytes that are close to the regions that are

394more exposed to their substrates (mainly methane and sulfide), since these compounds might be more easily

18 18 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

395accessible to symbionts first in the frontal zone compared to the abfrontal zone, sustaining bacterial growth.

396Among deleterious compounds, copper ions were for example recently hypothesized to cause cell apoptosis

397in Bathymodiolus azoricus. Cadmium was shown to cause apoptosis on isolated Crassostrea gigas

398hemocytes after 24h in vitro incubation in the range of 10-100 µmol.L-1 [61]. Gills of mussels are known to

399accumulate metals, and Bathymodiolus azoricus exposed to cadmium display antioxidant enzymatic

400activities that may be partly due to the endosymbionts. These could contribute to protect bacteriocytes, but

401not the symbiont-free ciliated cells, resulting in higher apoptosis rates in the latter [62]. In vivo incubation

402experiments at in situ pressure would be necessary to test these oxidative stress and fluid toxicity hypotheses.

403 Apoptosis of circulating and resident hemocytes have a high baseline level in mollusks [29], and we

404estimated that 34.5 % of the hemocytes were apoptotic in B. azoricus and B. puteoserpentis. The immune

405system of mussels relies on innate immunity only, and hemocytes play a key role in the immune response.

406They are also suspected to have a role in detoxification [58]. In Crassostrea virginica, the most common

407apoptotic cell type is the hemocyte. The apoptotic index for Crassostrea hemocytes, calculated the same way

408as herein, but using another visualization method (Apoptag), ranges between 23 and 99% with a mean of

40946% [63]. In another study, Crassostrea virginica displayed between 4.5% and 15.3% of apoptotic

410circulating hemocytes (Annexin-V assay) [64]. These results are congruent with the rates observed in this

411study, and could reveal overall high apoptosis rates in the hemocytes of bivalves, visually confirming the

412importance of apoptosis in the innate immune system.

413 Despite the fact that apoptosis rates are high in Bathymodiolus bacteriocytes compared to those in

414M. edulis epithelial gill-cells, they are only roughly half of those measured in ciliated cells and hemocytes.

415Results from transcriptomic studies and their subsequent interpretation by other authors led to the hypothesis

416that apoptosis was the main mechanism that allowed mussels (initially Bathymodiolus thermophilus) to

19 19 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

417control the amount of symbionts in their gills. The underlying idea was that bacteria-filled bacteriocytes

418would undergo apoptosis in a way similar to that observed in the tubeworm Riftia pachyptila, in which this

419mechanism participates to the recovery of symbionts carbon, and recycling of the animal cells [11]. On the

420contrary, our observations show that the bacteriocytes that contain only a few, or no bacteria at all, are

421mainly those that enter into apoptosis. Lower levels of apoptosis in bacteriocytes, in particular in those where

422symbionts are the most abundant, could be explained by the bacterial protection concerning the toxic

423compounds, which could provoke apoptosis of the cell. This could explain why cells with fewer bacteria,

424mostly located in the distal (i.e. abfrontal) edge away from the ciliated zone and possibly most depleted in

425symbiont substrates, would preferentially enter apoptosis, although this is merely hypothetical at this stage.

426In bacteriocytes near the ciliated zone, bacteria are most probably digested [21,23,65].

427 Overall, it appears that high rates of apoptosis are a normal feature in deep-sea mussel gills

428physiology. Rather than being a direct mechanism used to kill the most bacteria-rich cells in order to gain

429carbon, apoptosis seems to be involved in the overall dynamics of the gill organ itself. Indeed, observed

430patterns indicate that the cells harboring few to no bacteria are more often apoptotic. This can be interpreted

431in the light of gill specialization to symbiosis, with over-developed gills requiring stronger cilia activity,

432which possibly leads to higher turnover of ciliated cells, and habitat characteristics including the presence of

433toxic compounds. Apoptosis mechanisms are known to play major roles in other symbioses [4,5,11,66]. For

434example Codakia orbiculata, a shallow water bivalve that hosts Gammaproteobacteria in its gills can lose

435and reacquire symbionts. After loss of the symbionts, the bacteriocytes multiply massively to reacquire

436bacteria, while the cells in excess are eliminated by apoptosis [40]. A similar mechanism may be occurring in

437mussels. At first, apoptosis could appear as a heavy cost for the host, especially given that it may endanger

438the whole organ. However, neighboring cells could phagocytose the cells that undergo apoptosis, allowing

439recycling of their constituents. Apoptotic cells also tend to detach from the basal lamina, and once detached

20 20 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

440they may be treated as food particles by the mussels, again allowing recycling. The correlation observed in

441previous transcriptome analyses [25] between apoptotic rates and overall symbiont content may then be

442indirect: hosts with more bacteria tend to have larger gills [8], and thus more active ciliated cells, so overall

443more cells in apoptosis. High rates of apoptosis should thus be regarded as an integral part of the adaptation

444of deep-sea mussels to their symbioses and their habitats, and visualization pattern indicates that apoptosis

445probably plays more complex roles than previously assumed. This study has only investigated adult

446specimens, but because mussels acquire their symbionts in their post-larval stage and throughout the host

447lifetime [67,68], the coupling of symbiont acquisition with apoptosis should be explored at all developmental

448stages. Further study should also investigate the impact of stress (toxic compounds, temperature) and those

449of symbiont loss (when Bathymodiolus is moved away from the fluids compounds [18,69]) on the rates of

450apoptosis, as well as cell proliferation patterns. Because of the relative simplicity of the mussel holobiont, its

451high bacterial load and high levels of apoptosis, deep-sea mussels may prove useful models to further

452investigate the links between apoptosis, autophagy and symbiosis.

453

454Acknowledgments:

455We thank the captain and crew on board the Resarch Vessel “Pourquoi pas?” and the ROV “Victor 6000”

456(Ifremer) for their invaluable contributions to this work. A special thanks to chief scientists Dr. Karine Olu

457(WACS 2011), Pr. François Lallier (BioBaz 2013) and Dr. Marie-Anne Cambon-Bonavita and Dr. Magali

458Zbinden (BICOSE 2014). We are indebted to Dr. Kamil Szafranski for his help with the deep-sea mussels B.

459azoricus and B. puteoserpentis recovered on board during the BICOSE and BioBaz cruises. We are grateful

460to Antoine Mangin, Myriam Lebon and Marie Louvigné for their technical assistance as undergraduate

461students in the lab. This research was supported by Institut Universitaire de France project ACSYMB,

21 21 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

462Université Pierre et Marie Curie and ARED Région Bretagne project FlexSyBi (contract number 9127, grant

463to BP), BALIST program (ANR-08-BLAN-0252), EU projects EXOCET/D (FP6-GOCE-CT-2003-505342),

464MIDAS (FP7/2007-2013, grant agreement n◦ 603418) and CNRS and IFREMER for cruise funding. We are

465grateful to the microscopy platform Merimage (Roscoff, France) partner core facilities and to the Institute of

466Biology Paris-Seine Imaging Facility, supported by the “Conseil Regional Ile-de France”, CNRS and

467Sorbonne Université.

468

469Author’s contribution:

470B.P did the histology work, the microscopy imaging, the statistical analyses and wrote a first version of the

471paper. B.S. performed the isobaric mussel collections with the PERISCOP pressure-maintaining device

472during the two cruises (BioBaz 2013 and BICOSE 2014). F.L. enabled the collections of deep-sea mussel

473during the BioBaz 2013 cruise. S.D fixed the mussels during the WACS 2011 and BioBaz 2013 cruises and

474supervised BP’s analyses in the AMEX-team in Paris. AA fixed the M. edulis and contributed to BP’s

475histology work in the ABICE-team in Roscoff. S.D and AA are the collaborative designers of the study.

476S.D., A.A. and B.P. wrote the latest version of the paper, all authors have read and corrected its final draft.

477

478References:

479[1] Kerr JFR, Wyllie AH, Currie AR. Apoptosis: a basic biological phenomenon with wide-ranging 480 implications in tissue kinetics. Br J Cancer 1972;26:239–57. 481[2] Häcker G. The morphology of apoptosis. Cell Tissue Res 2000;301:5–17. doi:10.1007/s004410000193. 482[3] Elmore S. Apoptosis: a review of programmed cell death. Toxicol Pathol 2007;35:495–516. 483 doi:10.1080/01926230701320337.

22 22 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

484[4] Vigneron A, Masson F, Vallier A, Balmand S, Rey M, Vincent-Monégat C, et al. recycle 485 endosymbionts when the benefit is over. Curr Biol 2014;24:2267–73. doi:10.1016/j.cub.2014.07.065. 486[5] Dunn SR, Schnitzler CE, Weis VM. Apoptosis and autophagy as mechanisms of dinoflagellate symbiont 487 release during cnidarian bleaching: every which way you lose. Proc R Soc B Biol Sci 2007;274:3079–85. 488 doi:10.1098/rspb.2007.0711. 489[6] Dunn SR, Weis VM. Apoptosis as a post-phagocytic winnowing mechanism in a coral-dinoflagellate 490 mutualism. Environ Microbiol 2009;11:268–76. doi:10.1111/j.1462-2920.2008.01774.x. 491[7] Duperron S. The diversity of deep-sea mussels and their bacterial symbioses. In: Kiel S, editor. Vent 492 Seep Biota, vol. 33, Dordrecht: Springer; 2010, p. 137–67. 493[8] Duperron S, Quiles A, Szafranski KM, Léger N, Shillito B. Estimating symbiont abundances and gill 494 surface areas in specimens of the hydrothermal vent mussel Bathymodiolus puteoserpentis 495 maintained in pressure vessels. Front Mar Sci 2016;3:1–12. doi:10.3389/fmars.2016.00016. 496[9] Fisher CR, Childress JJ, Oremland RS, Bidigare RR. The importance of methane and thiosulfate in the 497 of the bacterial symbionts of two deep-sea mussels. Mar Biol 1987;96:59–71. 498[10] Sibuet M, Olu K. Biogeography, biodiversity and fluid dependence of deep-sea cold-seep communities 499 at active and passive margins. Deep Sea Res Part II 1998;45:517–67. doi:10.1016/S0967- 500 0645(97)00074-X. 501[11] Pflugfelder B, Cary SC, Bright M. Dynamics of cell proliferation and apoptosis reflect different life 502 strategies in hydrothermal vent and cold seep vestimentiferan tubeworms. Cell Tissue Res 503 2009;337:149–65. doi:10.1007/s00441-009-0811-0. 504[12] Lorion J, Kiel S, Faure B, Kawato M, Ho SYW, Marshall B, et al. Adaptive radiation of chemosymbiotic 505 deep-sea mussels. Proc R Soc B Biol Sci 2013;280:20131243–20131243. doi:10.1098/rspb.2013.1243. 506[13] Fiala-Médioni A, Métivier C, Herry A, Le Pennec M. Ultrastructure of the gill of the hydrothermal-vent 507 mytilid Bathymodiolus sp. Mar Biol 1986;92:65–72. doi:10.1007/BF00392747. 508[14] Halary S, Riou V, Gaill F, Boudier T, Duperron S. 3D FISH for the quantification of methane- and 509 sulphur-oxidizing endosymbionts in bacteriocytes of the hydrothermal vent mussel Bathymodiolus 510 azoricus. ISME J 2008;2:284–92. doi:10.1038/ismej.2008.3. 511[15] Szafranski KM, Piquet B, Shillito B, Lallier FH, Duperron S. Relative abundances of methane- and 512 sulfur-oxidizing symbionts in gills of the deep-sea hydrothermal vent mussel Bathymodiolus azoricus 513 under pressure. Deep Sea Res Part Oceanogr Res Pap 2015;101:7–13. doi:10.1016/j.dsr.2015.03.003.

514[16] Riou V, Halary S, Duperron S, Bouillon S, Elskens M, Bettencourt R, et al. Influence of CH4 and H2S 515 availability on symbiont distribution, carbon assimilation and transfer in the dual symbiotic vent 516 mussel Bathymodiolus azoricus. Biogeosciences 2008;5:1681–91. doi:10.5194/bg-5-1681-2008. 517[17] Guezi H, Boutet I, Andersen AC, Lallier FH, Tanguy A. Comparative analysis of symbiont ratios and 518 gene expression in natural populations of two Bathymodiolus mussel species. Symbiosis 2014;63:19– 519 29. doi:10.1007/s13199-014-0284-0. 520[18] Kádár E, Bettencourt R, Costa V, Santos RS, Lobo-da-Cunha A, Dando P. Experimentally induced 521 endosymbiont loss and re-acquirement in the hydrothermal vent bivalve Bathymodiolus azoricus. J 522 Exp Mar Biol Ecol 2005;318:99–110. doi:10.1016/j.jembe.2004.12.025. 523[19] Riou V, Colaço A, Bouillon S, Khripounoff A, Dando P, Mangion P, et al. Mixotrophy in the deep sea: a 524 dual endosymbiotic hydrothermal mytilid assimilates dissolved and particulate organic matter. Mar 525 Ecol Prog Ser 2010;405:187–201. doi:10.3354/meps08515. 526[20] Barry JP, Buck KR, Kochevar RK, Nelson DC, Fujiwara Y, Goffredi SK, et al. Methane-based symbiosis in 527 a mussel, Bathymodiolus platifrons, from cold seeps in Sagami Bay, Japan. Invertebr Biol 528 2002;121:47–54. 529[21] Fiala-Medioni A, Michalski J-C, Jollès J, Alonso C, Montreuil J. Lysosomic and lysozyme activities in the 530 gill of bivalves from deep hydrothermal vents. Comptes Rendus Académie Sci Paris 1994;317:239–44. 531[22] Streams ME, Fisher CR, Fiala-Medioni A. Methanotrophic symbiont location and fate of carbon 532 incorporated from methane in a hydrocarbon seep mussel. Mar Biol 1997;129:465–476. 533[23] Kádár E, Davis SA, Lobo-da-Cunha A. Cytoenzymatic investigation of intracellular digestion in the 534 symbiont-bearing hydrothermal bivalve Bathymodiolus azoricus. Mar Biol 2008;153:995–1004. 535 doi:10.1007/s00227-007-0872-0.

23 23 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

536[24] Boutet I, Ripp R, Lecompte O, Dossat C, Corre E, Tanguy A, et al. Conjugating effects of symbionts and 537 environmental factors on gene expression in deep-sea hydrothermal vent mussels. BMC Genomics 538 2011;12:1–13. doi:10.1186/1471-2164-12-530. 539[25] Guezi H, Boutet I, Tanguy A, Lallier FH. The potential implication of apoptosis in the control of 540 chemosynthetic symbionts in Bathymodiolus thermophilus. Fish Shellfish Immunol 2013;34:1709. 541 doi:10.1016/j.fsi.2013.03.223. 542[26] Bettencourt R, Pinheiro M, Egas C, Gomes P, Afonso M, Shank T, et al. High-throughput sequencing 543 and analysis of the gill tissue transcriptome from the deep-sea hydrothermal vent mussel 544 Bathymodiolus azoricus. BMC Genomics 2010;11:559. doi:10.1186/1471-2164-11-559. 545[27] Wong YH, Sun J, He LS, Chen LG, Qiu J-W, Qian P-Y. High-throughput transcriptome sequencing of the 546 cold seep mussel Bathymodiolus platifrons. Sci Rep 2015;5:16597. doi:10.1038/srep16597. 547[28] Zheng P, Wang M, Li C, Sun X, Wang X, Sun Y, et al. Insights into deep-sea adaptations and host- 548 symbiont interactions: a comparative transcriptome study on Bathymodiolus mussels and their 549 coastal relatives. Mol Ecol 2017:5133–48. doi:10.1111/mec.14160. 550[29] Sokolova IM. Apoptosis in molluscan immune defense. Invertebr Surviv J 2009;6:49–58. 551[30] Kiss T. Apoptosis and its functional significance in molluscs. Apoptosis 2010;15:313–21. 552 doi:10.1007/s10495-009-0446-3. 553[31] Romero A, Novoa B, Figueras A. The complexity of apoptotic cell death in mollusks: An update. Fish 554 Shellfish Immunol 2015;46:79–87. doi:10.1016/j.fsi.2015.03.038. 555[32] Barros I, Mendes S, Rosa D, Serrão Santos R, Bettencourt R. Vibrio diabolicus immunomodulatory 556 effects on Bathymodiolus azoricus during long-term acclimatization at atmospheric pressure. J Aquac 557 Res Dev 2016;7. doi:10.4172/2155-9546.1000464. 558[33] Cosel R von, Métivier B, Hashimoto J. Three new species of Bathymodiolus (Bivalvia: Mytilidae) from 559 hydrothermal vents in the Lau Basin and the north Fiji Basin, western Pacific, and the Snake Pit area, 560 Mid-Atlantic Ridge. The Veliger 1994;37:374–92. 561[34] Cosel R von, Comtet T, Krylova EM. Bathymodiolus (Bivalvia: Mytilidae) from hydrothermal vents on 562 the Azores Triple Junction and the Logatchev Hydrothermal Field, Mid-Atlantic Ridge. The Veliger 563 1999;42:218–48. 564[35] Olu-Le Roy K, Cosel R von, Hourdez S, Carney SL, Jollivet D. Amphi-Atlantic cold-seep Bathymodiolus 565 species complexes across the equatorial belt. Deep Sea Res Part Oceanogr Res Pap 2007;54:1890– 566 911. doi:10.1016/j.dsr.2007.07.004. 567[36] Fiala-Médioni A, McKiness Z, Dando P, Boulegue J, Mariotti A, Alayse-Danet A, et al. Ultrastructural, 568 biochemical, and immunological characterization of two populations of the mytilid mussel 569 Bathymodiolus azoricus from the Mid-Atlantic Ridge: evidence for a dual symbiosis. Mar Biol 570 2002;141:1035–43. doi:10.1007/s00227-002-0903-9. 571[37] Pradillon F. High hydrostatic pressure environments. In: Bell EM, editor. Life Extrem. Environ. Org. 572 Strateg. Surviv., Wallingford: CABI; 2012, p. 271–95. doi:10.1079/9781845938147.0271. 573[38] Pradillon F, Gaill F. Pressure and life: some biological strategies. Rev Environ Sci Biotechnol 574 2007;6:181–95. doi:10.1007/s11157-006-9111-2. 575[39] Shillito B, Hamel G, Duchi C, Cottin D, Sarrazin J, Sarradin P-M, et al. Live capture of megafauna from 576 2300m depth, using a newly designed Pressurized Recovery Device. Deep Sea Res Part Oceanogr Res 577 Pap 2008;55:881–9. doi:10.1016/j.dsr.2008.03.010. 578[40] Elisabeth NH, Gustave SDD, Gros O. Cell proliferation and apoptosis in gill filaments of the lucinid 579 Codakia orbiculata (Montagu, 1808) (Mollusca: Bivalvia) during bacterial decolonization and 580 recolonization. Microsc Res Tech 2012;75:1136–46. doi:10.1002/jemt.22041. 581[41] Gavrieli Y, Sherman Y, Ben-Sasson SA. Identification of programmed cell death in situ via specific 582 labeling of nuclear DNA fragmentation. J Cell Biol 1992;119:493–501. 583[42] Lallier F. BIOBAZ 2013 cruise, Pourquoi pas ? R/V 2013. doi:10.17600/13030030. 584[43] Cambon-Bonavita M-A. BICOSE cruise, Pourquoi pas ? R/V 2014. doi:10.17600/14000100. 585[44] Ravaux J, Hamel G, Zbinden M, Tasiemski AA, Boutet I, Léger N, et al. Thermal limit for metazoan life 586 in question: In vivo heat tolerance of the pompeii worm. PLoS ONE 2013;8:e64074. 587 doi:10.1371/journal.pone.0064074. 588[45] Olu K. WACS cruise, Pourquoi pas ? R/V 2011. doi:10.17600/11030010.

24 24 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

589[46] Abràmoff MD, Magalhães PJ, Ram SJ. Image processing with ImageJ. Biophotonics Int 2004;11:36–42. 590 doi:https://doi.org/10.1016/j.tcb.2004.03.002. 591[47] Duperron S, Guezi H, Gaudron SM, Pop Ristova P, WenzhöFer F, Boetius A. Relative abundances of 592 methane- and sulphur-oxidising symbionts in the gills of a cold seep mussel and link to their potential 593 energy sources: Variability of symbiont abundances in a seep mussel. Geobiology 2011;9:481–91. 594 doi:10.1111/j.1472-4669.2011.00300.x. 595[48] de Torres C, Munell F, Ferrer I, Reventos J, Macaya A. Identification of necrotic cell death by the 596 TUNEL assay in the hypoxic-ischemic neonatal rat brain. Neurosci Lett 1997;230:1–4. 597[49] Labat-Moleur F, Guillermet C, Lorimier P, Robert C, Lantuejoul S, Brambilla E, et al. TUNEL apoptotic 598 cell detection in tissue sections: critical evaluation and improvement. J Histochem Cytochem 599 1998;46:327–34. doi:10.1177/002215549804600306. 600[50] Kroemer G, Galluzzi L, Vandenabeele P, Abrams J, Alnemri ES, Baehrecke EH, et al. Classification of cell 601 death: recommendations of the Nomenclature Committee on Cell Death 2009. Cell Death Differ 602 2009;16:3–11. 603[51] Motta CM, Frezza V, Simoniello P. Caspase 3 in molluscan tissues: Localization and possible function. J 604 Exp Zool Part Ecol Genet Physiol 2013;319:548–59. doi:10.1002/jez.1817. 605[52] Boatright KM, Salvesen GS. Mechanisms of caspase activation. Curr Opin Cell Biol 2003;15:725–31. 606 doi:10.1016/j.ceb.2003.10.009. 607[53] Yuan J, Najafov A, Py BF. Roles of Caspases in Necrotic Cell Death. Cell 2016;167:1693–704. 608 doi:10.1016/j.cell.2016.11.047. 609[54] Chipuk JE, Green DR. Do inducers of apoptosis trigger caspase-independent cell death? Nat Rev Mol 610 Cell Biol 2005;6:268–75. doi:10.1038/nrm1573. 611[55] Abraham MC, Shaham S. Death without caspases, caspases without death. Trends Cell Biol 612 2004;14:184–93. doi:10.1016/j.tcb.2004.03.002. 613[56] Childress JJ, Barnes AT, Quetin LB, Robison BH. Thermally protecting cod ends for the recovery of 614 living deep-sea animals. Deep Sea Res 1978;25:419–22. doi:10.1016/0146-6291(78)90568-4. 615[57] Shillito B, Gaill F, Ravaux J. The IPOCAMP pressure incubator for deep-sea fauna. J Mar Sci Technol 616 2014;22:97–102. doi:10.6119/JMST-013-0718-3. 617[58] Fiala-Medioni A. Synthèse sur les adaptations structurales liées à la nutrition des mollusques bivalves 618 des sources hydrothermales profondes. Oceanol Acta 1988;N°SP:173–9. 619[59] Mone Y, Monnin D, Kremer N. The oxidative environment: a mediator of interspecies communication 620 that drives symbiosis evolution. Proc R Soc B Biol Sci 2014;281:20133112–20133112. 621 doi:10.1098/rspb.2013.3112. 622[60] Circu ML, Aw TY. Reactive oxygen species, cellular redox systems, and apoptosis. Free Radic Biol Med 623 2010;48:749–62. doi:10.1016/j.freeradbiomed.2009.12.022. 624[61] Sokolova IM. Cadmium-induced apoptosis in oyster hemocytes involves disturbance of cellular energy 625 balance but no mitochondrial permeability transition. J Exp Biol 2004;207:3369–80. 626 doi:10.1242/jeb.01152. 627[62] Company R, Serafim A, Bebianno MJ, Cosson R, Shillito B, Fiala-Médioni A. Effect of cadmium, copper 628 and mercury on antioxidant enzyme activities and lipid peroxidation in the gills of the hydrothermal 629 vent mussel Bathymodiolus azoricus. Mar Environ Res 2004;58:377–81. 630 doi:10.1016/j.marenvres.2004.03.083. 631[63] Sunila I, LaBanca J. Apoptosis in the pathogenesis of infectious diseases of the eastern oyster 632 Crassostrea virginica. Dis Aquat Organ 2003;56:163–70. doi:10.3354/dao056163. 633[64] Goedken M, Morsey B, Sunila I, Dungan C, De Guise S. The effects of temperature and salinity on 634 apoptosis of Crassostrea virginica hemocytes and perkinsus marinus. J Shellfish Res 2005;24:177–83. 635 doi:10.2983/0730-8000(2005)24[177:TEOTAS]2.0.CO;2. 636[65] Détrée C, Chabenat A, Lallier FH, Satoh N, Shoguchi E, Tanguy A, et al. Multiple I-type lysozymes in the 637 hydrothermal vent mussel Bathymodiolus azoricus and their role in symbiotic plasticity. PLOS ONE 638 2016;11:1–19. doi:10.1371/journal.pone.0148988. 639[66] Foster JS, McFall-Ngai MJ. Induction of apoptosis by cooperative bacteria in the morphogenesis of 640 host epithelial tissues. Dev Genes Evol 1998;208:295–303.

25 25 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

641[67] Wentrup C, Wendeberg A, Huang JY, Borowski C, Dubilier N. Shift from widespread symbiont infection 642 of host tissues to specific colonization of gills in juvenile deep-sea mussels. ISME J 2013;7:1244–7. 643[68] Wentrup C, Wendeberg A, Schimak M, Borowski C, Dubilier N. Forever competent: deep-sea bivalves 644 are colonized by their chemosynthetic symbionts throughout their lifetime: Symbiont colonization in 645 gills of deep-sea bivalves. Environ Microbiol 2014;16:3699–713. doi:10.1111/1462-2920.12597. 646[69] Raulfs EC, Macko SA, Van Dover CL. Tissue and symbiont condition of mussels (Bathymodiolus 647 thermophilus) exposed to varying levels of hydrothermal activity. J Mar Biol Assoc UK 2004;84:229– 648 34. doi:10.1017/S0025315404009087h. 649[70] Le Pennec M, Hily A. Anatomie, structure et ultrastructure de la branchie d’un Mytilidae des sites 650 hydrothermaux du pacifique oriental. Oceanol Acta 1984;7:517–23. 651

26 26 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

653Figure 1

654

27 27 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

655Figure 2

656

28 28 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

658Figure 3

659

29 29 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

660

661

662

663

664Figure 4

665

30 30 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

667Figure 5

668

31 31 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

669Figure 6

32 32 bioRxiv preprint doi: https://doi.org/10.1101/422840; this version posted September 21, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

671Figure 7

33 33